首先是下载，下载官方的二进制包， 下载地址 （支持Linux、Darwin、Freebsd和Windows）。哈哈：

 wget https://s3.amazonaws.com/bitly-downloads/nsq/nsq-1.0.0-compat.linux-amd64.go1.8.tar.gz

解压：

tar -xzf nsq-1.0.0-compat.linux-amd64.go1.8.tar.gz

然后建立目录：

mkdir -p /home/vagrant/nsq-1.0.0-compat.linux-amd64.go1.8/bin/config
cd /home/vagrant/nsq-1.0.0-compat.linux-amd64.go1.8/bin/config

建立配置文件dmin.cnf,内容如下：

## log verbosity level: debug, info, warn, error, or fatal
log-level = "info"
## <addr>:<port> to listen on for HTTP clients
http_address = "0.0.0.0:4171"
## graphite HTTP address
graphite_url = ""
## proxy HTTP requests to graphite
proxy_graphite = false
## prefix used for keys sent to statsd (%s for host replacement, must match nsqd)
statsd_prefix = "nsq.%s"
## format of statsd counter stats
statsd_counter_format = "stats.counters.%s.count"
## format of statsd gauge stats
statsd_gauge_format = "stats.gauges.%s"
## time interval nsqd is configured to push to statsd (must match nsqd)
statsd_interval = "60s"
## HTTP endpoint (fully qualified) to which POST notifications of admin actions will be sent
notification_http_endpoint = ""
## nsqlookupd HTTP addresses
nsqlookupd_http_addresses = [
    "127.0.0.1:4161"
        ]
## nsqd HTTP addresses (optional)
#nsqd_http_addresses = [
#    "127.0.0.1:4151"
#]

建立nsqd.cnf,内容如下：

## log verbosity level: debug, info, warn, error, or fatal
log-level = "info"
## unique identifier (int) for this worker (will default to a hash of hostname)
# id = 5150
## <addr>:<port> to listen on for TCP clients
tcp_address = "0.0.0.0:4150"
## <addr>:<port> to listen on for HTTP clients
http_address = "0.0.0.0:4151"
## <addr>:<port> to listen on for HTTPS clients
# https_address = "0.0.0.0:4152"
## address that will be registered with lookupd (defaults to the OS hostname)
# broadcast_address = ""
## cluster of nsqlookupd TCP addresses
nsqlookupd_tcp_addresses = [
    "127.0.0.1:4160"
]
## duration to wait before HTTP client connection timeout
http_client_connect_timeout = "2s"
## duration to wait before HTTP client request timeout
http_client_request_timeout = "5s"
## path to store disk-backed messages
# data_path = "/var/lib/nsq"
## number of messages to keep in memory (per topic/channel)
mem_queue_size = 10000
## number of bytes per diskqueue file before rolling
max_bytes_per_file = 104857600
## number of messages per diskqueue fsync
sync_every = 2500
## duration of time per diskqueue fsync (time.Duration)
sync_timeout = "2s"
## duration to wait before auto-requeing a message
msg_timeout = "60s"
## maximum duration before a message will timeout
max_msg_timeout = "15m"
## maximum size of a single message in bytes
max_msg_size = 1024768
## maximum requeuing timeout for a message
max_req_timeout = "1h"
## maximum size of a single command body
max_body_size = 5123840
## maximum client configurable duration of time between client heartbeats
max_heartbeat_interval = "60s"
## maximum RDY count for a client
max_rdy_count = 2500
## maximum client configurable size (in bytes) for a client output buffer
max_output_buffer_size = 65536
## maximum client configurable duration of time between flushing to a client (time.Duration)
max_output_buffer_timeout = "1s"
## UDP <addr>:<port> of a statsd daemon for pushing stats
# statsd_address = "127.0.0.1:8125"
## prefix used for keys sent to statsd (%s for host replacement)
statsd_prefix = "nsq.%s"
## duration between pushing to statsd (time.Duration)
statsd_interval = "60s"
## toggle sending memory and GC stats to statsd
statsd_mem_stats = true
## message processing time percentiles to keep track of (float)
e2e_processing_latency_percentiles = [
    1.0,
    0.99,
    0.95
]
## calculate end to end latency quantiles for this duration of time (time.Duration)
e2e_processing_latency_window_time = "10m"
## path to certificate file
tls_cert = ""
## path to private key file
tls_key = ""
## set policy on client certificate (require - client must provide certificate,
##  require-verify - client must provide verifiable signed certificate)
# tls_client_auth_policy = "require-verify"
## set custom root Certificate Authority
# tls_root_ca_file = ""
## require client TLS upgrades
tls_required = false
## minimum TLS version ("ssl3.0", "tls1.0," "tls1.1", "tls1.2")
tls_min_version = ""
## enable deflate feature negotiation (client compression)
deflate = true
## max deflate compression level a client can negotiate (> values == > nsqd CPU usage)
max_deflate_level = 6
## enable snappy feature negotiation (client compression)
snappy = true

建立lookupd.cnf, 内容如下：

## log verbosity level: debug, info, warn, error, or fatal
log-level = "info"
## <addr>:<port> to listen on for TCP clients
tcp_address = "0.0.0.0:4160"
## <addr>:<port> to listen on for HTTP clients
http_address = "0.0.0.0:4161"
## address that will be registered with lookupd (defaults to the OS hostname)
# broadcast_address = ""
## duration of time a producer will remain in the active list since its last ping
inactive_producer_timeout = "300s"
## duration of time a producer will remain tombstoned if registration remains
tombstone_lifetime = "45s"

然后运行如下命令：

mkdir -p /home/vagrant/nsqlog
cd /home/vagrant/nsq-1.0.0-compat.linux-amd64.go1.8/bin
nohup ./nsqd -config=/home/vagrant/nsq-1.0.0-compat.linux-amd64.go1.8/bin/config/nsqd.cnf >> /home/vagrant/nsqlog/nsqd.log 2>&1 &
nohup ./nsqlookupd -config= /home/vagrant/nsq-1.0.0-compat.linux-amd64.go1.8/bin/config/lookupd.cnf >> /home/vagrant/nsqlog/nsqlookupd.log 2>&1 &
nohup ./nsqadmin -config=/home/vagrant/nsq-1.0.0-compat.linux-amd64.go1.8/bin/config/admin.cnf >> /home/vagrant/nsqlog/nsqadmin.log 2>&1 &

成功。至此，可以登录http://192.168.33.10:4171 查看后台了。
如此简单。
这是配置
参考：http://ju.outofmemory.cn/entry/271510